Question

I try to activate libstemmer for a website.
My sphinx version is Sphinx 0.9.9-release (r2117) on a Debian server, and everything is working fine in my dev env (Mac OSX, same sphinx).

I have this error while indexing :

invalid morphology option 'libstemmer_french' - IGNORED

My options are

min_prefix_len = 4
enable_star = 1
html_strip = 1
charset_table = (...)
morphology = libstemmer_french

After reading more about libstemmer, I suspect that my sphinx install does not support libstemmer.
It is a production server so I can’t afford to try and fail too much. My questions are :

• How can I check if libstemmer has been compiled with sphinx ?
• How can I install libstemmer above sphinx / or reinstall sphinx in a safe way ?

Note : I installed sphinx using :

apt-get install sphinx

…so no idea of the configure options.

Answer 1

Sphinx by default works only with Russian and English stemming. You have to compile sphinx with another version libstemmer_c to get support for french stemming:

Please, follow these steps:

1) Get the last sphinx version:

svn checkout http://sphinxsearch.googlecode.com/svn/trunk/ sphinxsearch-read-only

2) Get the libstemmer_c library:

wget http://snowball.tartarus.org/dist/libstemmer_c.tgz

3) Replace libstemmer_c in sphinx source directory:

tar xf libstemmer_c.tgz && cp -rfp libstemmer_c/* sphinxsearch-read-only/libstemmer_c/

4) Configure sphinx with appropriate parameters:

./configure --enable-id64 --with-libstemmer && make